Learning from missing data with the binary latent block model

Missing data can be informative. Ignoring this information lead to misleading conclusions when the model does not allow extracted from missing data. We propose a co-clustering model, based on binary Latent Block Model, that aims take advantage of nonignorable nonresponses, also known as Not At Random A variational expectation–maximization algorithm is derived perform inference and selection criterion presented. assess proposed approach simulation study, before using our voting records lower house French Parliament, where analysis brings out relevant groups MPs texts, together with sensible interpretation behavior non-voters.